Apple has filed a lawsuit against OpenAI in federal court in California, alleging the company orchestrated the theft of trade secrets through former employees. Names in the filing include Tang Tan, former VP of product design for iPhone and Apple Watch, and Chang Liu, senior electrical engineer.

The complaint asserts that Tan used Apple’s internal code names during interviews with candidates still employed at Apple, asking them to bring actual hardware components to meetings. Apple claims to have collected evidence of confidential file downloads by candidates hours before OpenAI interviews, and found an internal Apple document on exit security protocols distributed by Tan to new OpenAI hires before they left Apple.

Why it matters. If you work at a company developing technology or collaborating with AI vendors, this lawsuit sets a precedent: it defines what constitutes “theft” when knowledge moves with people. AI partnerships that seemed solid can close quickly if one party suspects the other is drawing on confidential information through targeted hiring. For those building with large language model APIs, it means monitoring terms of service and confidentiality clauses: what you share today could end up in a competitor’s product tomorrow.

In detail

Context

The lawsuit comes as OpenAI is building a hardware business in collaboration with Jony Ive, Apple’s former chief design officer. OpenAI acquired io Products, Ive’s startup, in a $6.5 billion deal in 2025, bringing over 50 engineers and designers along. Ive had founded io with Scott Cannon, Evans Hankey (who led Apple’s design team after Ive’s departure), and Tang Tan himself. Neither Ive, Hankey, nor Cannon are personally named in the initial filing.

Apple says it raised the issue directly with OpenAI in February 2026, requesting an internal investigation. OpenAI never responded. The complaint describes what has emerged so far as “the tip of the iceberg,” arguing Apple lacks visibility into what actually happens inside OpenAI, where this behavior would allegedly be normalized at the top.

What Apple alleges in detail

According to the filing, Tan allegedly used confidential Apple project code names to probe candidates still employed at Apple, explicitly asking “what’s the plan?” about unreleased products. He also reportedly instructed candidates to bring “actual” Apple hardware parts to meetings for “show and tell” sessions, surprising at least one candidate who commented “I didn’t even know we could take them out of the office.”

OpenAI allegedly requested candidates provide “CAD artifacts and designs,” “prototypes,” and details on component selection, integration tools, and supplier relationships. Apple claims to have found a pattern: employees preparing to leave for OpenAI downloading confidential files shortly before interviews. In one documented case, a candidate started taking screenshots and downloading materials on a highly confidential Apple project hours before meeting Tan, who then asked for more information about that same project during the interview.

Apple also states that Tan distributed an internal document called “Need to Know” on exit security protocols to new OpenAI hires before they resigned from Apple, supposedly to help them evade controls.

What it means for AI partnerships

The lawsuit calls into question the soundness of collaborations between companies building hardware and those training models. Apple has a partnership with OpenAI announced in 2024 to integrate ChatGPT into certain system features, but that collaboration now unfolds as the two companies face off in court. If Apple obtains an injunction limiting OpenAI’s use of confidential information, it could slow or block development of the company’s hardware products.

For those using commercial model APIs, the practical lesson is verifying what happens to data you pass through. Terms of service typically exclude using your inputs to train public models, but don’t always cover what happens if someone from your organization moves to the other side with knowledge you shared internally. Non-compete and confidentiality clauses are difficult to enforce in the United States, especially in California, where it’s nearly impossible to prevent someone from working for a competitor. This lawsuit tries to shift focus from the employee to the company hiring them, arguing OpenAI orchestrated the theft. If Apple wins, it becomes riskier to hire from a competitor without rigorous checks on what the person brings with them.

What we don’t know yet

Apple explicitly states it lacks complete visibility into what was taken and how it was used. The lawsuit relies on what the company could reconstruct through file access audits and candidate testimony. We don’t know if OpenAI actually used the information in its product design, nor whether employees beyond those named are involved. OpenAI has not yet responded publicly to the lawsuit at the time of publication.